An in-memory data grid (IMDG) consists of multiple interconnected virtual machines, referred to as Java® Virtual Machines (JVMs) (Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates), which provide an address space for locating data for ease of access by applications. In one example, a database or other data source may be loaded into memory to an IMDG. By hosting the data in-memory, application performance is enhanced, since demand for physical server resources, particularly Input/Output (I/O) overhead to physical disk devices is reduced. In another example, an IMDG may be used to provide transient storage for data that is created, for example, within a web application session. Depending on the application's requirements, different IMDG partitioning schemas may be chosen. However, reconfiguring the IMDG to maintain performance levels or to respond to changing data access patterns is a disruptive manual process that may require extensive system unavailability. If an autonomic IMDG partitioning mechanism can be implemented based on data usage, then system availability and high performance levels may be maintained.